Extracting an image from a rect

[RenatoEsposito1999]

Hi everyone.
I'm having a hard time extracting an image from a rectangle.

If I have the coordinates that allow me to correctly draw the area I'm interested in on the PDF, why can't I get the image with get_pixmap using the same coordinates?

For example (rect given as input to the function):

rect = fitz.Rect(rect)
zoom = dpi / 72  # 72 dpi = 1.0
mat = fitz.Matrix(zoom, zoom)
pix = page.get_pixmap(matrix=mat, clip=clip, annots=False)
out_path="some_paths.png"
pix.save(out_path)

The images it manages to save are of completely messed up areas, then at a certain point I get:

return _mupdf.fz_save_pixmap_as_png(pixmap, filename)
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
pymupdf.mupdf.FzErrorArgument: code=4: Invalid bandwriter header dimensions/setup

[JorjMcKie]

Please provide the problem file / page.
BTW: You are not extracting an image at all. Your code renders a sub-area (the clip) of that page.
You do not say what you mean by "messed up areas" - potentially the page is rotated or whatever.

[RenatoEsposito1999]

Unfortunately, I can't share files. I'll try to explain better.
A rectangle drawn on a PDF has these coordinates:
DRAW RECT: Rect(96.9000015258789, 1126.9520263671875, 116.70000457763672, 1146.7520751953125) (this is a print(rect))
If I try to extract a pixmap with those coordinates, the code fails:

return _mupdf.fz_save_pixmap_as_png(pixmap, filename)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
pymupdf.mupdf.FzErrorArgument: code=4: Invalid bandwriter header dimensions/setup

For another rectangle:
DRAW RECT: Rect(0.47998046875, 0.15234375, 841.3800048828125, 1190.7322998046875)
The code doesn't fail, but the image isn't the area I expect.
Perhaps the coordinates are being interpreted differently; do I need to do something with them?
The portion of the code is:

zoom = dpi / 72
mat = fitz.Matrix(zoom, zoom)
pix = page.get_pixmap(matrix=mat, clip=rect)

width = page.rect.width
height = page.rect.height

give me:

width: 1191.0
height: 842.0

[JorjMcKie]

Your code could be simpler, but otherwise looks correct.
Again: you extract nothing but rather make a photo from a subarea of the page.
How this may go wrong I cannot tell without having the file. Is the page rotated?

You could could simply do:

pix = page.get_pixmap(clip=rect, dpi=150)
pix.save("output.png")

If you data protection concerns you can send the file to my email address instead of attaching it here. In Artifex, we are all bound by Non Disclosure Agreements, so rest assured that your data is safe.

[RenatoEsposito1999]

Unfortunately, I have a file restriction order.
I noticed that the image is rotated 90°, but the draw_Rect is correct. Could the get_pixmap have problems?
(The original file is rotated.)

[RenatoEsposito1999]

Problem solved.
Before processing the page, I set the rotation to 0 and everything works fine.

[JorjMcKie]

I cannot say anything else to your case because I have no facts (i.e. a file or page) at hand.
And I don't want to hallucinate ... as I am human and not an AI 😅

[RenatoEsposito1999]

Sorry about that.
But the problem was the rotations, the page was rotated 90°.

[JorjMcKie]

no worries